Jump to content
Sign in to follow this  
denoir

OFP Community Census Project

Recommended Posts

I think you need to add more than Bush, Not Bush or Don't care as this is a very generalized statement. Also, people can believe in God and not believe in Religion. Maybe you should change it to Religious, Not Religious, Or Spiritual but not Religious.

Share this post


Link to post
Share on other sites

I don't really understand the cluster pictures, but the resulting statistics are very interesting. Especially since they don't count at all for me smile_o.gif

Share this post


Link to post
Share on other sites
I don't really understand the cluster pictures, but the resulting statistics are very interesting. Especially since they don't count at all for me  smile_o.gif

It's not very complicated. Clusters are simply a grouping by similarity.

Imagine this: You ask people two questions, for instance their age and their income. You then get a 2d plot looking something like this (age on the x-axis and income on the y-axis - the dots are the data points):

cluster3.jpg

As you can see, there are two separate groups. A clustering algorithm finds thos groups:

cluster3a.jpg

Now should we ask three questions - i.e have three variables, we would get a 3d plot:

cluster3b.jpg

To find the clusters, you simply measure the distance between the data points. Things that are close to each other form a cluster. Things that are far away belong to another cluster. Hence similiarty is just a question of distance between the data points.

In the case of our survey we have more variables which makes it impossible to plot. We have 9 variables, and a 9d plot is a bit difficult to visualize. That's why we have the 9 different plots for each variable. For instance:

cluster3c.jpg

It's quite simple - they're color coded. You see the color scale and it matches to an age. Blue areas are younger people while the red ones are the older people.

The black dots are represent data points (single or overlapping) that are sorted by similarity - i.e the distance between the points in the 9-dimensional space. The boundaries between groups of data points - i.e clusters are marked by the black line. For looking at just the clusters, we look at the "Clusters" plot:

cluster3d.jpg

In the example above we have separated the data into two clusters.

Bottom line, the clusters represent groupings by similarity in the data. When we use data from our survey we get groupings by 'types' of people.

Share this post


Link to post
Share on other sites
People under 18 years old

Below average post count.

Strongly above average warning levels.

Strongly above average postive to Bush (positive to him)

Strongly below average negative to EU as a superpower (negative to it)

Strongly above average religious.

Well only the post count is correct here (570 or so).

Don't have any warning levels, I'm negative towards Bush, positive towards the EU as a superpower and not at all religious...

So I guess I am a freak sad_o.gifcrazy_o.giftounge_o.gif

Share this post


Link to post
Share on other sites
People under 18 years old

Below average post count.

Strongly above average warning levels.

Strongly above average postive to Bush (positive to him)

Strongly below average negative to EU as a superpower (negative to it)

Strongly above average religious.

Well only the post count is correct here (570 or so).

Don't have any warning levels, I'm negative towards Bush, positive towards the EU as a superpower and not at all religious...

So I guess I am a freak  sad_o.gif  crazy_o.gif  tounge_o.gif

Relative your age group average - yes  wink_o.gif

You have to keep in mind that we're talking about averages here. There will always be individual cases that differ from the average. Also worth remembering is that we're still working with pretty limited data (47 submissions so far), so you can't draw any broader conclusions yet. When we get to about 100 data points, then we can perhaps start saying that it is representative for the OFP forum members.

Given a total population of 20,000 (approx number of forum members) and 47 replies, we have a confidence of +-14%. So, there is a fuzzy border between the clusters.

Share this post


Link to post
Share on other sites

...hey denoir, i was not loged in but fill every field and submited, does it count since i put my nick ?, and where i can see most resent results ?

btw.... good idea but litle bit confuse at the start  tounge_o.gif  biggrin_o.gif

Share this post


Link to post
Share on other sites
...hey denoir, i was not loged in but fill every field and submited, does it count since i put my nick ?, and where i can see most resent results ?

btw.... good idea but litle bit confuse at the start  tounge_o.gif  biggrin_o.gif

Your data was recorded. And you can see the most recent results when I post them in this thread smile_o.gif

Share this post


Link to post
Share on other sites

wow, very interesting results

I math the old-age above average postcount uncommitted European image fairly well

thanks for explaing the data mining, I start liking the off-topic forum more biggrin_o.gif

Share this post


Link to post
Share on other sites

Hi All

Data mining is a great tool for buisiness I used a Seagates Data mining tool to plot the effects of ticket pricing on sales at a climbing wall I worked for. It allowed us to totaly remodel the pricing structure to concentrate on cheap annual memberships. This changed many clients who visited the wall irregularly in to long term members.

The wall went from just surviving, to being able to double in size and profitability in 3 months and tripple in size in 9 months.

The pricing structure was aimed at multiple catagories of clients based on age, usage, income, employment etc. By setting up the structure correctly we were able set up each different layer in the pricing of period so that it always made more sense to the client to buy a longer period ticket.

Since all the prices, with the exception of single visit tickets where its price remained the same, were actualy lower than the previous prices clients could not believe their luck.

It is now the most succesfull wall in the UK with thousands of members visiting every day. All the other walls in the UK have started following the same price structure.

Kind Regards Walker

Share this post


Link to post
Share on other sites

If you check the forum rules you should notice there is currently no ruling saying you must participate rock.gifsmile_o.gif

Share this post


Link to post
Share on other sites

Fine next time I´ll put in where the jokes are etc.

[supporting]Well I did it just to help him out, it looks like a really neet project that if he worked in marketing then it would be super. It´s a great idea, now I can say to peoples comments: Offcorse you say so your from that country biggrin_o.gif [/supporting]

[explaining]The thing about Bush, it was just a joke Im not here to trash people, but you got to admit if he did this on milion americans Bush would pay big money for it. [/explaining]

Share this post


Link to post
Share on other sites
Fine next time I´ll put in where the jokes are etc.

All it takes is one of these ---> wink_o.gif

smile_o.gif

Share this post


Link to post
Share on other sites
I think you need to add more than Bush, Not Bush or Don't care as this is a very generalized statement. Also, people can believe in God and not believe in Religion. Maybe you should change it to Religious, Not Religious, Or Spiritual but not Religious.

stop ignoring us denoir.

Share this post


Link to post
Share on other sites
I think you need to add more than Bush, Not Bush or Don't care as this is a very generalized statement. Also, people can believe in God and not believe in Religion. Maybe you should change it to Religious, Not Religious, Or Spiritual but not Religious.

stop ignoring us denoir.

I'm not ignoring you - I think you are right - the religion question could have been more nuanced . But it's too late to change things now, for this survey - lot of people have already responded.

Share this post


Link to post
Share on other sites

Anyway, before I'm off to bed, the latest data:

cluster4.jpg

For an explanation of the color values, see:

http://www.flashpoint1985.com/cgi-bin....1;st=42

The clusters:

cluster4b.jpg

Cluster 1 - marked as red in the "Clusters" plot

CLICK FOR LARGE IMAGE OF CLUSTER

People with spiritual beliefs, againt Bush and for the EU becoming a superpower.

Percent of population: 8.6%

Age - Slightly above average.

Origin - Mostly Europeans, but others too.

Post Count - Above average

Warning levels - Above average

Bush - All against.

EU superpower - All for.

Religious - All have spiritual beliefs.

The people in this cluster:

Quote[/b] ]

Hellfish6

Placebo

Radnik

python3

CrackerJack

Cluster2 - marked as blue  in the "Clusters" plot

CLICK FOR LARGE IMAGE OF CLUSTER

Non-religious younger Europeans against Bush and for an EU superpower.

Percent of population: 25.9%

Age - Slightly below average.

Origin - Mostly Europeans.

Post Count - Below average

Bush - Strongly against.

EU superpower - All for.

Religious - Are not religious.

The people in this cluster:

Quote[/b] ]

toadlife

Jakerod

Tovarish

Architekt

SmartDrv[Kyllikki]

Turms

CosmicCastaway

Pathy

D.murphy man

Gollum1

FW200

Gadger

MonkeyGoat

Xawery

[ZG]BUZZARD

Cluster3 - marked as yellow in the "Clusters" plot

CLICK FOR LARGE IMAGE OF CLUSTER

Percent of population: 13.8%

Non-religious older Europeans with higher post counts, against Bush and for an EU superpower.

Age - Above average.

Origin - All Europeans.

Post Count - Above average

Bush - All against.

EU superpower - Strongly for.

Religious - Are not religious.

The people in this cluster:

Quote[/b] ]

Joltan

denoir

Balschoiw

Ran

walker

Killswitch

redface

Lt_Phoenix

Cluster4 - marked as orange in the "Clusters" plot

CLICK FOR LARGE IMAGE OF CLUSTER

Europeans against Bush and against EU as a superpower.

Percent of population: 32.7%

Age - Below average.

Origin - Mostly Europeans.

Post Count - Below average.

Bush - Strongly against.

EU superpower - Strongly against.

Religious - Mixed.

The people in this cluster:

Quote[/b] ]

EiZei

$33ker

GoOB

Ironsight

Tymi

Kopijeger

KevBaz

Llauma

hardrock

da12thMonkey

HotShot

Status_cz

ozanzac

R. Gerschwarzenge

iNeo

MunkD

King Homer

STGN

der bastker

Cluster5 - marked as red in the "Clusters" plot

CLICK FOR LARGE IMAGE OF CLUSTER

The pro-Bush camp. Mostly Americans. Mostly not caring or against the EU as a superpower.

Percent of population: 18.9%

Age - Below average.

Origin - Mostly Americans but with a significant minority of Europeans.

Post Count - Below average.

Bush - Strongly for.

EU superpower -  Neutral (above average against).

Religious - Mixed. (Above average religious)

The people in this cluster:

Quote[/b] ]

radkeff

SgtBarnes

blackdog

Viper200

MSMS_KDXer

shinraiden

oyman

billybob2002

General Barron

rudedog

DeadMeat

-------------------

One thing that I perhaps should have mentioned - the clustering is performed by a neural network type called a "self-organizing map". It autonomously finds the clusters, so there is no human bias in the clustering.

Share this post


Link to post
Share on other sites

Unfortunately you made a bad generalization about Bush, I am against Bush, but I want him to win now just for the people who support him.

Odd eh, I'll say no.

Share this post


Link to post
Share on other sites

oh, i read something wrong in your census. i read "Do you think that the EU should become a military superpower?" as "Do you think that the EU becomes a military superpower?"

crazy_o.gifghostface.gif

Share this post


Link to post
Share on other sites
Quote[/b] ]Unfortunately you made a bad generalization about Bush, I am against Bush, but I want him to win now just for the people who support him.

Odd eh, I'll say no.

"18.9%" of the population... tounge_o.giftounge_o.gif It is better than I thought... wink_o.gif ...expect it to go down some.... tounge_o.gif

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  

×